On 18/06/15 20:07, Yair Yarom wrote:

> We solved it a bit differently. Under the assumption that a program
> should specify how much memory it requires, instead of getting a random
> amount according to which node/partition it's running on, we set
> DefMemPerCPU = 50 for the entire cluster(s).

That's just what we do too, we set our default not based on hardware,
but looking back at our Torque logs pre-migration and increasing from
the 1GB per core then to 2GB/core with DefMemPerCPU=2048 (default is 1
CPU core too).

If you want more you need to ask for it.

The benefit of that is that for us is that as our current Intel clusters
have minimums of 4GB/core and 16GB/core then if a lot of people can get
away inside that limit then you can still fit larger jobs into the
memory and cores left over.

All the best,
Chris
-- 
 Christopher Samuel        Senior Systems Administrator
 VLSCI - Victorian Life Sciences Computation Initiative
 Email: [email protected] Phone: +61 (0)3 903 55545
 http://www.vlsci.org.au/      http://twitter.com/vlsci

Reply via email to